DACT

DACT performs composite-null hypothesis testing to detect DNA methylation-mediated effects of exposures on clinical outcomes in genome-wide epigenetic studies.

Key Features:

• Composite Null Hypothesis Testing: Evaluates composite null hypotheses when assessing mediation effects across numerous genomic sites.
• Empirical Null Framework: Implements Efron's empirical null framework to estimate the null distribution under composite-null conditions.
• Comparison to Traditional Tests: Accounts for limitations of Wald-type Sobel’s test and joint significance tests that assume standard null distributions and suffer power loss in finite samples.
• Improved Power and Type I Error Control: Demonstrates increased power to detect mediation effects while controlling the type I error rate in finite samples.
• Epigenome-wide Application: Targets genome-wide DNA methylation data to identify mediating CpG sites of exposure effects on clinical outcomes.
• Sensitivity and Robustness Analysis: Includes sensitivity analysis for unmeasured confounding with reported residual correlations of error terms between outcome and mediator regressions < 0.01.

Scientific Applications:

• Epigenome-wide mediation analysis: Applied to large-scale epigenetic datasets to discover DNA methylation mediators of exposure–outcome relationships.
• Normative Aging Study application: Identified multiple CpG sites mediating the effect of smoking on lung function (effect sizes −0.18 to −0.79) with a controlled false discovery rate at 0.05, highlighting genes such as AHRR and F2RL3.

Methodology:

Performs composite-null hypothesis testing using Efron's empirical null framework, compares performance to Wald-type Sobel’s test and joint significance tests, controls type I error and false discovery rate, and conducts sensitivity analysis via residual correlation assessment between outcome and mediator regression errors.